I'm working on a project which tests participants' perception of how coherent they think two sentences are. They would judge on a scale of 1-7 (7=most coherent while 1=least coherent).
Participants read 4 different kinds of sentence structures
and each structure is put into a story. Each story has 4 pairs of sentences and they are of different levels of coherence. There are 10 stories totally for each kind of sentence structures. In other words, participant will read 40 sets of stories, each set contains 4 pair of sentences, and they make judgement for each pair.
My hypothesis is for some sentence structures, there will be greater variance than others but it depends on the language proficiency of the participants. I have now collected the frequency of each rate (1-7) for each kind of sentences. I try to work on the variance and use an anova repeated measure to see if the variance gets larger in some kinds of sentences. Yet I'm not sure if it is the right way to do so. I'm also wondering if I can measure how much the sentence structures would influence of degree of variance and see how it interacts (correlates) with the level of proficiency as I have two test scores of their language proficiency.
I'm sorry for the long post.
Advertise on Talk Stats